If you follow the link posted by user1153046, it gives normal instructions for installing under Linux.

For WSL, it turns out you don't need to install specific GPU drivers, per official NVidia Documentation, just add the TensorFlow libraries/binaries/dependencies for Cuda with one pip install line:

pip install tensorflow[and-cuda]

Which installs these additional packages:

Installing collected packages: nvidia-nvjitlink-cu12, nvidia-nccl-cu12, nvidia-curand-cu12, nvidia-cufft-cu12, nvidia-cuda-runtime-cu12, nvidia-cuda-nvrtc-cu12, nvidia-cuda-nvcc-cu12, nvidia-cuda-cupti-cu12, nvidia-cublas-cu12, nvidia-cusparse-cu12, nvidia-cudnn-cu12, nvidia-cusolver-cu12

Previously, I installed tensorflow without the "[and-cuda]" and got the same error message: "Could not find cuda drivers on your machine, GPU will not be used."